THE emergence of OpenAI’s ChatGPT over the past year has demonstrated the vast potential of Artificial Intelligence (AI).
For businesses, however, public AI’s like ChatGPT represents an opportunity and a challenge. It can solve a lot of problems but could also result in sensitive information being potentially leaking out.
Deloitte Africa’s AI and Data Director, Dr. Quentin Williams, tells us why Generative AI (an AI model that is capable of generating text, images, and other media) is a good fit for businesses and that they should consider incorporating one of their own into their operations.
CBN: Will Generative AI practice allow businesses to create the equivalent of ChatGPT for their operations?
QW: Yes definitely. Generative AI allows you to take the foundation models that ChatGPT uses or that of Google or AWS and fine tune it with the business’ own data and for their own business purposes.
Examples of this are the creation of document Q&As, ChatBots and querying your enterprise data, but within a private space.
This can, therefore, create a model that is specific to their “way of thinking” using their own templates, brand guidelines and their own intellectual property. We have seen businesses like Bloomberg and NVIDIA doing this.
CBN: If so, why create it? Why not just use ChatGPT?
QW: ChatGPT is broad based and will provide generative content that is very general and not tailored to or aligned with the business’ own data, IP, brand guidelines or templates.
They can also set this up in a private space, where the business data will not be shared with the outside world.
A business specific fine-tuned model can then also be used for multiple purposes. This provides a new level of differentiation for a business when they are dealing with their customers or trying to provide operational efficiencies.
CBN: ChatGPT is a remarkable tool, but is there a danger that businesses will only look at what it is capable of and limit their thinking of what AI is capable of?
QW: Indeed. ChaptGPT is a conversational bot based on Generative AI.
Generative AI, and indeed AI, is a much larger field that includes machine learning, data analysis and robotics.
Apart from generating content, Generative AI can also be used to do analyses, recommendations, modelling large unstructured data and incorporating document-based (text or image or any kind of data) data into various structured forms and databases.
This allows us nowadays to combine traditional AI with Generative AI to create new applications such as customer behaviour analyses and engagement, and even be used to create advertising content adjustment based on customer preferences.
It can also be used to make real-time operational adjustments to improve efficiencies, like generating a new factory layout or production schedule on the fly.
CBN: Can you provide an example (a case study is preferable) on how Generative AI has transformed a business in Africa?
QW: Generative AI is fairly new, and thus there aren’t many examples where it has yet transformed a business.
In Africa though, we do have case studies where generative AI has allowed us to create new businesses and new innovations.
Examples of these are Lelapa.AI which is creating new context-specific, African language AI products to serve the African market, allowing business to engage with their customers more easily in local languages.
Similarly, African start-up InstaDeep has created a product called DeepChain that is transforming the way vaccine and protein discovery is done.